On 09.01.2015 17:25, Thad Guidry wrote:
https://www.wikidata.org/wiki/Property:P279 aka "the superclass" ... seems to have an equivalent property that refers to http://www.w3.org/2000/01/rdf-schema#subClassOf ???
Basically yes, this was the informal design intention when the community discussed these properties. However, we should be very careful when stating this as an "equivalent property" to rdfs:subClassOf, since the latter is not just some arbitrary vocabulary (like most other properties) but part of the RDFS and OWL language definitions that come with several possible semantics. Stating that our property is "equivalent" to such a language feature is not the whole story.
On a related note, it might really be preferrable to use properties "subClassOf(ExternalClass)" and "subPropertyOf(ExternalProperty)" to relate Wikidata elements to external vocabulary. Using these properties would mean that, from our viewpoint as a community, our claims are still correct if we would be using certain externally defined vocabulary elements instead of our own classes/properties.
== (Invented) Example ==
If we would have the claims
(1) Q1 P31 Q2 (2) Q2 subClassOf http://example.org/external/class/uri
then it means that we think that following claim is also meaningful:
(3) Q1 P31 http://example.org/external/class/uri
Our claim (2) tells external users that they may get instances of http://example.org/external/class/uri from our Q2 instances.
For this application, subClassOf (and similarly subPropertyOf) is enough. What you get with equivalentClass (equivalentProperty) is merely the other direction.
== Example (ctnd) ==
In the example above, if we would use
(2') Q2 equivalentClass http://example.org/external/class/uri
this would mean that we think that any instance of the class http://example.org/external/class/uri should also be in our Q2. It is a bit like an virtual import of data that we have not even seen. I doubt that we really want this.
The advantage of avoiding "equivalentProperty" in favour of "subPropertyOfExternalProperty" (and similarly for classes) is also that our definition can be more specific than the one used for the external property/class. So we can relate our data to multiple external things, which may fit our own definition more or less tightly.
Cheers,
Markus